DIVERSITY AMONG UNMARRIED PARENTS: HUMAN CAPITAL ... · unmarried parents, significant differences...
Transcript of DIVERSITY AMONG UNMARRIED PARENTS: HUMAN CAPITAL ... · unmarried parents, significant differences...
DIVERSITY AMONG UNMARRIED PARENTS: HUMAN CAPITAL,
ATTITUDES, AND RELATIONSHIP QUALITY
Center for Research on Child Wellbeing
Working Paper #02-01-FF
REVISED October 2002 August 2002
February 2002
Cynthia Osborne
DRAFT 9/2/02
Diversity Among Unmarried Parents:
The Importance of Marriage Expectations
Cynthia Osborne [email protected]
Princeton University Office of Population Research
Center for Research on Child Wellbeing
September 3, 2002
Working Paper
The Fragile Families and Child Wellbeing Study is supported by the following federal grants: 5R01-HD-35301 from the National Institute of Child Health and Human Development and 5P30-HD-32030 from the National Institute of Child Health and Human Development through the Office of Population Research, Princeton University. Additional funding provided by: California HealthCare Foundation; The Center for Research on Religion and Urban Civil Society at the University of Pennsylvania; Commonwealth Fund; Ford Foundation; Foundation for Child Development; Fund for New Jersey; William T. Grant Foundation; Healthcare Foundation of New Jersey; William and Flora Hewlett Foundation; Hogg Foundation; Christian A. Johnson Endeavor Foundation; Kronkosky Charitable Foundation; Leon Lowenstein Foundation; John D. and Catherine T. MacArthur Foundation; A.L. Mailman Family Foundation; Charles Stewart Mott Foundation; National ScienceFoundation; David and Lucile Packard Foundation; Public Policy Institute of California; Robert Wood Johnson Foundation; St. David's Hospital Foundation; St. Vincent Hospital and Health Services; U.S. Department of Health and Human Services (ASPE and ACF).
1
DRAFT 9/2/02
Abstract
As policy makers seek to devise programs to promote healthy marriages among
unmarried parents, significant differences between married and unmarried parents as well as
differences among unmarried parents must be taken into account. A majority of unmarried
parents express a desire to marry, but many obstacles exist to forming stable unions.
This paper uses data from the Fragile Families and Child Wellbeing survey to answer the
following questions: How different are unmarried parents from married parents in terms of their
demographic characteristics, human capital, economic resources, attitudes about marriage and
gender roles, and relationship quality? How do unmarried parents differ among themselves on
these domains? How important are marriage expectations in differentiating unmarried parents?
And to what extent do differences in demographic characteristics and socioeconomic status
explain differences in attitudes and relationship quality?
The results suggest that expectations of marriage are a key indicator of a couple’s
marriage potential. Comparing married parents to unmarried parents, based on their living
arrangement alone, provides incomplete information regarding the differences in human capital,
attitudes, and relationship quality. Cohabitors and visitors are quite similar on most domains.
However, mothers with high expectations of marriage have attitudes more favorable for marriage
and better relationship quality than do mothers with low marriage intentions, even after
controlling for their higher human capital.
These results have important implications in targeting effective policies to help stabilize
fragile families.
2
DRAFT 9/2/02
As a result of fewer women in marriages and declines in marital fertility, nonmarital
childbearing in the United States has increased significantly as a proportion of all births.
Whereas in 1970, 11 percent of children were born to unmarried mothers, in 1998, nearly a third
of all births were to unmarried women (National Center for Health Statistics, table 9). This
varies considerably by race, such that almost 70 percent of African American children are born
out-of-wedlock, compared to 22 percent of non-Hispanic white children (National Center for
Health Statistics, table 9). This is a concern given the negative consequences generally
associated with children who do not live with both of their biological parents, and the high rates
of poverty and public assistance use among single-parent families (McLanahan and Bumpass,
1988; Duncan and Rogers, 1991; McLanahan and Sandefur, 1994).
A significant portion of the decline or delay in marriages can be attributed to increases in
cohabitation. Premarital cohabitation has become the norm among adults, with more than half of
all first unions in the early 1990s beginning as cohabitation (Bumpass and Lu, 2000). Nearly
two-thirds of the decline in marriage by age 25 between 1970 and 1985 can be attributed to the
increase in cohabitation (Bumpass, Sweet, and Cherlin, 1991).
Cohabiting relationships involving children are increasing significantly, which is
represented by the growing proportion of nonmarital births to couples in cohabiting
relationships. In 1989, almost a quarter of nonmarital births were to cohabiting parents
(Bumpass and Sweet, 1989). By the mid 1990s the share of nonmarital births to cohabiting
parents had grown to almost 40 percent (Bumpass and Lu, 2000), and by 2000, half of all
nonmarital births were to cohabiting parents, most of whom plan to marry in the future
(McLanahan and Garfinkel, 2002).
3
DRAFT 9/2/02
The increase in children born to cohabiting parents is concerning given the instability and
lower economic resources associated with cohabiting unions. Relationships that begin as
cohabitation are shorter term and less stable than marriages, even if they progress to marriage
(Bumpass and Lu, 2000; Graefe and Lichter, 1999). Moreover, cohabitation is selective of
individuals with lower human capital (Bumpass and Lu, 2000; Manning and Lichter, 1996), and
cohabiting couples with higher economic resources tend to marry, particularly with the presence
of children (Manning and Smock, 1995; Smock and Manning, 1997), leaving those remaining in
cohabiting unions with lower economic resources.
Another nontrivial portion of nonmarital births are to couples who are romantically
involved but not living together when their child is born, referred to as visitors (McLanahan and
Garfinkel, 2002, Sigle-Rushton and McLanahan, 2001). McLanahan and her colleagues
identified that one-third of nonmarital births in large cities are to couples in a visiting
relationship, most of whom plan to marry or live together in the future and jointly raise their
child (McLanahan and Garfinkel, 2002). The nature of visiting relationships and the capacities
of the partners are less well defined than cohabiting relationships, due to lack of adequate data on
this group.
Although marriage is desirable for many of these unmarried couples, many obstacles
exist to acting on their plans (Sigle-Rushton and McLanahan, 2001). Identifying the constraints
on the capacities of unmarried couples and the differences in attitudes and relationship quality
between married and unmarried couples helps clarify the nature of these unmarried relationships
as well as the level of effort policy makers will have to employ as they devise programs to
promote marriage or stability among these fragile families.
4
DRAFT 9/2/02
This paper provides new information on unmarried parents in relation to married parents
and looks at the diversity among unmarried parents. I extend previous research on unmarried
parents in three ways. First, I examine a much richer set of individual and couple indicators,
including attitudes, capabilities, and relationship quality. Secondly, I examine couples in visiting
relationships who have received very little attention due to data limitations. Thirdly, I look at
differences among unmarried parents, taking into consideration both their living arrangement and
expectations of marriage. I extend the research on cohabitation by focusing exclusively on new
parents and by seeing whether the diversity found among cohabiting partnerships applies to new
parents as well.
In the next section of the paper I discuss previous research on unmarried couples,
focusing mainly on cohabiting couples and the importance of marriage intentions in
differentiating cohabiting couples. In the third section I describe the data, variables, and
methods used in my analysis. In the fourth section I discuss the results, and in the final section I
draw conclusions and discuss policy implications.
PREVIOUS STUDIES
Previous research on the family has focused mainly on comparing the economic
resources and child outcomes of married, divorced, and single-parent families, providing a clear
understanding that married parents have substantially more resources than unmarried families,
which contribute significantly to differences in child outcomes (McLanahan and Sandefur,
1994).
More recently, with the decline in marriage and rise in cohabitation, researchers have
begun to look extensively at cohabiting couples in relation to married couples, attempting to
5
DRAFT 9/2/02
discern if cohabitation is an alternative to marriage or a stage in the marriage process (Smock,
2000). Two general conclusions arise from the research on cohabiting couples. First, cohabitors
differ significantly from married couples and singles in many ways (Smock, 2000), and
secondly, cohabiting couples differ significantly from each other (Brown and Booth, 1996;
Casper and Sayer, 2000).
Cohabiting couples are likely to be less educated and have lower incomes than married
couples (Bumpass and Lu, 2000; Nock, 1995; Smock, 2000). They are also likely to categorize
themselves as liberal versus conservative, have more egalitarian gender roles, and be less
religious than married couples (Clarkberg, Stolzenberg, and Waite, 1995; Smock, 2000).
Cohabiting couples also report less satisfaction and happiness with their relationships and lower
relationship quality overall, which may contribute to the greater instability of cohabiting couples
compared to married partners (Nock, 1995; Brown and Booth, 1996). When cohabitors are
compared to both married couples and singles in terms of socioeconomic characteristics and
views on marriage, they fall between the two, but are generally more similar to singles than
married couples (Rindfus and VandenHuevel, 1990).
Despite the differences between married and cohabiting couples and singles, it is
important not to view cohabiting couples as a monolithic group. Casper and Sayer (2000)
identified four distinct types of cohabiting couples based on attitudes about marriage and
intentions to marry. Some cohabiting couples consider cohabitation a substitute for marriage.
They usually have rather stable relationships, but these relationships persist as cohabitation and
are not likely to transition into marriage. Another group of cohabitors views cohabitation as a
precursor to marriage. These couples are the most likely to transition to marriage and their
attitudes are most similar to married couples. Two other cohabiting groups are those that use
6
DRAFT 9/2/02
cohabitation as a trial period for marriage and those that see it as an alternative to steady dating.
Each of these unions is likely to dissolve and not result in marriage or continued cohabitation. In
general, they find that intentions or expectations to marry are highly correlated with behavior.
Other research confirms the importance of expectations of marriage in differentiating
unmarried groups and their transition to marriage (Brown and Booth, 1996; Brown 2000; Waller
and McLanahan, 2001). Brown and Booth (1996) found that although cohabitors reported lower
relationship quality than married couples overall, these differences could be largely explained by
expectations for marriage. Most cohabitors expect to marry, and those with high marriage
expectations have similar relationship quality as married couples. Brown (2000) shows that
cohabitors’ relationship assessment and expectations of marriage are positively linked with union
transition, while Waller and McLanahan (2001) find that while mutual expectations for marriage
and cohabitation are both important in predicting marriage among unmarried parents, marriage
expectations have a larger effect than living arrangement.
Marriage intentions are a declaration of being in the marriage market, therefore, it is
understandable that this group would be more likely to marry, and that this variable would
differentiate unmarried couples along other domains as well. Marriage intentions are not always
mutual between partners (Bumpass, Sweet, and Cherlin, 1991, Brown and Booth, 1996, Brown
2000), but even with partner disagreement, the results change little (Brown and Booth, 1996).
Despite all that has been learned recently about cohabiting couples, two gaps remain in
the current research. Little is known about cohabiting couples with children in relation to
married parents (see Manning and Lichter, 1996 as an exception), and virtually no research has
looked at romantically involved couples that are trying to jointly raise a child while not living
7
DRAFT 9/2/02
together (identified as visitors in this paper), to determine if the diversity in cohabiting
relationships applies to visitors as well.
In this paper, I focus exclusively on new parents. Further, I compare visiting as well as
cohabiting parents to married parents; and I differentiate unmarried parents by their living
arrangement and marriage expectations. Understanding the differences between married and
unmarried parents, and the differences among unmarried parents can help identify the nature of
the relationships of unmarried parents and help target policies to help unmarried families become
more stable.
I predict that married couples will have higher human capital, attitudes more favorable
for marriage, and better relationship quality than cohabitors or visitors. However, among the
unmarried, it is unclear if cohabiting and visiting parents with high marriage expectations will be
more similar to married parents, or if cohabitors, regardless of marriage intentions, will be more
similar to married parents especially in their attitudes and relationship quality, given the higher
level commitment generally associated with cohabitation.
DATA
This analysis uses newly available data from the Fragile Families and Child Well-Being
Study. This data set allows for a comparison of cohabiting parents to married parents with a new
child. It also provides previously unavailable data on parents who are romantically involved, but
not living together when the child is born, referred to as visitors. Other data sets generally
categorize women who are not married or cohabiting as single and not involved in a relationship
with the baby’s father. However, the Fragile Families study shows that visiting relationships
8
DRAFT 9/2/02
make up a third of nonmarital births (Sigle-Rushton and McLanahan, 2002), and therefore
should be considered in analyses relating to unmarried families.
The Fragile Families study collected baseline data in twenty cities throughout the United
States from 1998 to 2000, and is nationally representative of cities of more than 200,000 people.
A cohort of married and unmarried mothers (and most fathers) was interviewed in the hospital
shortly after the birth of the child, and the parents will be reinterviewed when the child is 12
months, 30 months, and 60 months.
This paper uses baseline information from 4221 married and unmarried mothers in all
twenty cities. The sample is limited to mothers who report having a romantic relationship with
the baby’s father to understand the role marriage expectations play in distinguishing unmarried
parents. This excludes 651 mothers who were not in a romantic relationship with the baby’s
father at the child’s birth and an additional 18 mothers that did not respond to the question
regarding their expectations of marrying the baby’s father.
Variables and Methodology
The analysis begins with a comparison of the demographic characteristics, human capital,
attitudes about marriage, attitudes about gender roles, and relationship quality between married
and unmarried parents, and among unmarried parents. Subsequently, I determine the extent to
which differences in demographic characteristics, socioeconomic status and marriage
expectations explain the differences in eight attitude and relationship quality variables. Lastly, I
use the demographic characteristics and socioeconomic variables to create predicted probabilities
of the eight attitude and relationship quality measures for each relationship type.
Attitudes about marriage are assessed using three scales (pro marriage for the couple, pro
marriage for the child, and marriage as a financial institution) based on a composite of the
9
DRAFT 9/2/02
mother’s responses to various interview questions for each attitude. The pro marriage for the
couple scale is comprised of the mother’s responses to two questions: 1)“It is better for a couple
to get married than just live together” and 2)“Living together is just the same as being married”
(opposite coded). Responses ranged from 1 (strongly disagree) to 4 (strongly agree). The pro
marriage for the child scale was based on the two questions: 1)“A single mother can bring up a
child as well as a married couple” (opposite coded) and 2)“It is better for children if their parents
are married.” Each scale has a maximum score of eight and a minimum of two. The marriage as
a financial institution scale was comprised of three questions: 1) “How important do you think
the husband having a steady job is for a successful marriage?” 2) “How important do you think
the wife having a steady job is for a successful marriage?” and 3) “The main advantage of
marriage is financial security.” The first two questions were recoded so that 3 equaled very
important and 1 equaled not important. The third question allowed for a 1 (strongly disagree) to
4 (strongly agree) response. I recoded it on a three point scale, collapsing agree (3) and strongly
agree (4) into one category. The maximum score for this scale is nine and the minimum is three.
Two scales measure the mother’s attitudes about gender roles: one to measure traditional
gender role views, and one to measure gender distrust. Each scale is based on two questions
with a 1 (strongly disagree) to 4 (strongly agree) response, and a maximum score of eight and
minimum of two. The traditional gender role scale is comprised of the two questions: 1)“The
important decisions in the family should be made by the man,” and 2)“ It is better if the husband
earns the living and the woman cares for the family.” The gender trust scale is based on the two
questions 1)“In dating, a man is largely out to take advantage of a woman,” and 2)“Men cannot
be trusted to be faithful,” such that a high score represents low levels of trust.
10
DRAFT 9/2/02
Finally, to measure the mother’s relationship quality with the baby’s father, I developed
two scales, and I use a dichotomous variable to measure physical abuse. One scale is based on
the mother’s report of disagreement with the father about money, spending time together, sex,
the pregnancy, drugs/alcohol, and being faithful. The responses range from 1 (often) to 3 (never)
for a maximum score of 18 and minimum of six, with a high score indicating high levels of
agreement. The second relationship quality scale measures the degree of support the mother
feels from the father. The three questions in this scale include: 1)“Baby’s father is fair and
willing to compromise,” 2)“Baby’s father expresses affection or love for you,” and 3)“Baby’s
father encourages you or helps you do things that are important to you.” Each of the questions is
based on a 1 to 3 scale, recoded so that 3 represents often, for a total of nine and minimum of
three. Physical abuse is measured by the mother’s response to the question “Does the baby’s
father hit or slap you?” The 1 (often) to 3 (never) response was recoded to create a dummy
variable such that 0 equals no abuse and 1 equals any abuse.
The mother was supposed to be alone during the interview, however the baby’s father
and other relatives could have been nearby in the hospital. The preceding questions were also
asked directly after she had had a child with the man she is referring to. Therefore, these
responses might be underreporting actual levels of disagreement and abuse in the relationship.
The primary independent variables are the mother’s relationship status with the baby’s
father at the child’s birth and her expectations of marriage. Mothers are distinguished between
married (n=1186) and unmarried (n=3035) based on her report. All unmarried mothers in this
sample are in a romantic relationship with the baby’s father. Unmarried mothers who report they
are living with the baby’s father are classified as cohabiting (n=1774) and the other mothers are
considered to be in a visiting relationship (n=1261).
11
DRAFT 9/2/02
To understand the role expectations play in differentiating unmarried parents, I divided
the unmarried parents into four groups based on their living arrangement (cohabiting or visiting)
and expectations of marriage (high or low). Expectations of marriage represent the mother’s
reported chances of marriage, based on her response to the question “What do you think the
chances are that you will marry the baby’s father in the future.” Five responses were possible:
no chance, little chance, a 50-50 chance, a good chance, and an almost certain chance. A high
chance represents a response of good or almost certain, and low represents a response of a 50-50
chance or less. The four unmarried groups include: cohabiting/high (n=1390), cohabiting/low
(n=384), visiting/high (n=619), and visiting/low (n=642).
It is possible that unmarried mothers were overly optimistic about their chances of
marriage given that this question was asked shortly after their child’s birth, yet this should result
in fewer differences between the groups being identified.
Demographic characteristics and measures of human capital and employment that might
affect a mother’s attitudes are also included in the models. I created four dummy variables to
represent mother’s race including, non-Hispanic white (referred to as white), non-Hispanic black
(referred to as black), Hispanic, and other. Nativity is a dummy variable coded as one if the
mother reports being born outside of the United States. Age is specified as a continuous
variable. Religiosity is a dummy variable coded as one if the mother reports weekly attendance
at a religious service.
Human capital includes the mother’s and father’s employment, education, mother’s self-
reported health status, father’s health problems that limit work, and payment of the birth by
Medicaid to capture low-income status. Employment for the mother is a dummy variable,
defined as having earnings from work in the year prior to the birth of the child. Father
12
DRAFT 9/2/02
employment is also a dummy variable, defined as the father working for wages within the week
prior to the interview. Education is a set of four dummy variables including less than high
school, high school, some college or technical training, and college education or more. Health
status refers to the mother’s assessment of her current health status based on a four-point scale
ranging from excellent to poor. Fair and poor responses were combined to create three health
categories, and a dummy variable was created for each. Father’s health is a dichotomous
variable based on the mother’s report of any health problems that limit his ability to work.
Medicaid payment is also a dummy variable coded as one if the mother reported that Medicaid
was the only method of payment for the child’s birth. Father’s employment and education are
based on the father’s report where available, and supplemented with the mother’s report for 659
observations.
For each of the eight attitudinal and relationship quality variables, I ran three logistic
regression models to predict the difference between married and unmarried mothers in scoring
“high” (defined below) on the variable. The first model compares cohabiting and visiting
mothers to married mothers, without any controls. The second model adds in variables for
mothers’ age, race, nativity, religiosity, education, employment, Medicaid use and health, and
fathers’ employment. Only mother’s information (with the exception of employment) is
included in the models due to the high correlation between mother’s and father’s information.
The third model interacts living arrangement with marriage expectations, and compares the four
groups to married mothers.
A comparison of the coefficients of cohabitors and visitors between the first and second
models illustrates the extent to which differences in demographic characteristics and
socioeconomic status explain differences in attitudes and relationship quality. By comparing the
13
DRAFT 9/2/02
coefficients of the demographic and human capital covariates between the second and third
models I can determine the extent to which expectations have an independent effect on attitudes
and relationship quality. If marriage expectations are a reflection of being prepared to marry, I
would expect the coefficients on education and employment to decrease in size and significance
when expectations are introduced into the model.
Scoring “high” is defined as scoring 6 out of 8 on the pro marriage for the couple and
child scales and the two attitudes about gender roles scales; 9 out of 9 on the marriage as a
financial institution scale and supportiveness scale; scoring 1 versus 0 for physical abuse; and 9
or less out of 18 on the disagreement scale. In this case, the respondents have to disagree “often”
(1) about three of the areas and at least “sometimes” (2) about the others.
The goal of this analysis is to show the association between relationship status and
attitudes and relationship quality, net of differences in background; the goal is not to determine a
causal link. The parents in this sample selected into their respective relationships and living
arrangements based on many of the demographic and human capital characteristics being
analyzed in this paper (Carlson, et al, 2001). Carlson and her colleagues (2001) found that black
women are less likely to choose a coresidential relationship, and that education and father’s
employment increase the likelihood a couple will be in a cohabiting relationship or marriage.
Further, more favorable attitudes toward marriage, and more emotional support from the father
increases the likelihood new parents will be in a romantic relationship and that they will be
cohabiting or married.
14
DRAFT 9/2/02
RESULTS AND DISCUSSION
Differences in Demographic Characteristics and Human Capital
Married versus Unmarried Parents
Understanding the differences between married and unmarried parents provides insight
into the type and amount of resources unmarried parents may need in order to form stable
unions. Significant differences in human capital characteristics would suggest that investments
in policies related to education, job training, tax relief, child care, and health insurance would be
necessary. Differences in attitudes about marriage and gender relations, and in relationship
quality, especially net of capacity differentials would suggest that marriage counseling might be
necessary to help unmarried parents.
Table 1 lists the demographic and human capital characteristics of unmarried parents in
relation to married parents. The second column in the table combines cohabiting and visiting
parents to illustrate all unmarried parents in this sample. The results show considerable
differences between married and unmarried parents on all demographic and human capital
domains. Married parents in this sample are more likely to be white, while over half of the
unmarried parents are black. Married mothers are also more likely than unmarried mothers to be
born outside of the United States. Married parents are, on average, five years older than
unmarried parents, but the age difference between the mother and father is the same for married
and unmarried parents. Married mothers are also twice as likely to report attending a religious
service every week.
Employment differences between married and unmarried fathers are substantial. Over 90
percent of married fathers were working just before the baby was born, compared to 75 percent
15
DRAFT 9/2/02
of unmarried fathers. Employment differences between married and unmarried mothers are
smaller, but significant.
Education is the biggest area of difference between married and unmarried parents.
Approximately a third of married parents has a college degree or higher, compared to three
percent of unmarried parents. Two-fifths of unmarried parents in this sample did not graduate
from high school. Married mothers are also more likely to report being in excellent health, and
married fathers are half as likely to have a health problem that limits their ability to work.
Cohabiting versus Visiting Parents
Policy makers often assume that cohabiting couples are more committed and perhaps
more marriage ready than other romantically involved couples who are not living together. If
this is the case, cohabitors may need less help to create stable unions, or alternatively, policy
makers might want to focus on cohabitors since they would be an easier target.
These results show that cohabitors are not significantly different from visiting parents on
most domains. On average, cohabiting parents fall between married and visiting parents, but the
largest differences are between married and unmarried parents rather than among cohabiting and
visiting parents (see Table 1).
Table 1 shows the demographic and human capital characteristics of cohabiting and
visiting parents in relation to each other and to married parents. Cohabitors and visitors differ
largely by race and nativity. Approximately 70 percent of parents in visiting relationships are
black compared to about 45 percent of cohabiting parents in this sample. Over a third of
cohabiting mothers are Hispanic, compared to 20 percent of visitors, and 17 percent of
cohabiting mothers are foreign born compared to 10 percent of visiting mothers.
16
DRAFT 9/2/02
The two groups are more similar on the human capital domains, yet differences,
particularly in employment, exist. Cohabitors are more likely than visitors to be employed,
especially the fathers. Approximately 80 percent of cohabiting fathers were working prior to the
baby’s birth compared to two-thirds of visiting fathers. Almost 69 percent of cohabiting mothers
worked in the year prior to the baby’s birth, compared to 62 percent of mothers in a visiting
relationship. Cohabitors are, on average, slightly more educated than visitors. However, few
unmarried mothers or fathers in either group are highly educated, and again, this is where
unmarried parents differ most from married parents. Mothers in visiting relationships report
slightly better health than cohabiting mothers, and cohabiting fathers are somewhat more likely
than visiting fathers to have a health problem that limits his ability to work.
A significant area of difference between cohabiting and visiting mothers is their
expectations of marriage to the baby’s father. Almost 80 percent of cohabiting mothers report
high expectations of marriage compared to about half of visiting mothers. This difference is
large given the similarities these mothers share in other domains related to marriage readiness.
Marriage Expectations
Table 2 shows descriptive statistics for the four unmarried groups, based on their living
arrangement and expectations of marriage, in relation to each other and to married parents. This
shows that comparing cohabiting and visiting parents without regard to their marriage
expectations does not provide the full story. As mentioned in the first section of this paper,
marriage expectations have been shown to distinguish cohabiting couples in terms of their
commitment and similarities to married couples (Brown, 2000; Casper and Sayer, 2000). This
analysis confirms the previous studies and finds that this is also true for parents in visiting
17
DRAFT 9/2/02
relationships. Cohabiting and visiting parents with high marriage expectations are more similar
to each other and to married parents, than are other unmarried parents on most domains.
Within cohabiting and visiting relationships, those with high marriage expectations have
higher human capital and look more like parents who are married than do parents with low
marriage expectations. For example, cohabitors are, on average, more likely to be white than
visitors, yet cohabitors and visitors with high marriage expectations are more likely to be white
than are their counterparts with low marriage expectations. This holds true for religiosity, and
mother’s and father’s employment as well. Those with high marriage expectations report higher
levels of education and health than do parents with low marriage expectations, regardless of
living arrangement.
Differences in Attitudes and Relationship Quality
Model 1 in tables 3 through 5 shows that there are significant differences in attitudes and
relationship quality between married and unmarried mothers. The question remains, do the
significant differences in background characteristics and marriage expectations, described above,
explain these differences? Models 2 and 3 answer this question.
Attitudes about Marriage
Table 3 shows the results of scoring high on the three measures of attitudes about
marriage. Model 1 shows that in comparison to married mothers, cohabiting and visiting
mothers have significantly less favorable attitudes about marriage for the couple and the child,
and are more likely to think of marriage as a financial institution.
When background and socioeconomic characteristics are controlled (Model 2), the
differences between the groups are diminished, but remain large. Blacks are most likely to think
that marriage is the best institution for a couple, and the least likely to think it is best for a child.
18
DRAFT 9/2/02
Education is associated with more favorable attitudes about marriage, and thinking less of the
financial benefits of marriage. Father employment is associated with being more likely to think
that marriage is good for the couple, while mother employment is associated with being less
likely to think marriage is important for the child.
Model 2 also shows that cohabitors have slightly more favorable attitudes about marriage
than do visitors, and that they are much less likely to think of marriage as a financial institution.
All of the differences are significant at the .10 level or below.
Model 3 interacts living arrangement with marriage expectations and compares the four
groups of unmarried mothers to married mothers. This shows that mothers with high marriage
expectations as compared to low have more favorable attitudes about marriage for the couple and
child. Visitors are more likely to think of marriage as a financial institution, regardless of
marriage expectations.
If marriage expectations are a reflection of a mother’s preparation for marriage, then
looking at variables associated with marriage readiness would be sufficient, and looking at
marriage expectations would be unnecessary. However, if expectations predict attitudes about
marriage independent of human capital and demographic characteristics associated with marriage
readiness, then expectations are a useful tool in sorting out unmarried couples.
Indeed, model 3 shows that marriage expectations have an independent effect on
predicting attitudes about marriage. Comparing the coefficients on the covariates between model
2 and 3 shows little change in the odds ratios on the variables that are associated with marriage
readiness when expectations are included in the model.
19
DRAFT 9/2/02
Attitudes about Gender Roles
Unmarried mothers report less traditional views on gender roles than do married mothers,
and report significantly higher levels of distrust in men (see Table 4). Controlling for differences
in background characteristics does little to explain differences in attitudes about gender roles,
and explains about a quarter of the difference between married and unmarried mothers’ trust in
men. Older mothers are more traditional and less trusting. Mothers born outside of the United
States report less trust in men, as do black mothers. Black and white mothers hold similar views
on gender roles, and mothers who regularly attend religious services have twice the odds of
having traditional views. Education is linked with more trust in men and less traditional views,
as is mother’s employment. Use of Medicaid is linked with less trust in men, while health is
associated with greater trust.
Cohabiting and visiting mothers have similar attitudes about gender roles and levels of
trust in men. However, dividing these unmarried mothers by their expectations of marriage
shows a more complete picture. All unmarried mothers report similar views on gender roles,
while those with low marriage expectations are the ones with very low levels of trust in men.
Mothers with high marriage expectations report levels of trust similar to married mothers.
Again, marriage expectations seem to have an independent affect on attitudes about gender roles.
Expectations moderate some of the effect of race on gender distrust, but the differences by race
remain large and significant.
Relationship Quality
Table 5 shows that married mothers as compared to unmarried mothers report lower
levels of disagreement and physical abuse in their relationships with the baby’s father (model 1).
20
DRAFT 9/2/02
Married mothers report similar levels of emotional support from the baby’s father as cohabiting
mothers, but about twice the support of visiting mothers.
However, differences in background characteristics explain much of these differences.
Model 2 controls for differences in demographic and human capital characteristics and shows
that when these differences are taken into account cohabitors have similar levels of disagreement
as married mothers, there is no difference between married mothers and unmarried mothers in
terms of physical abuse in the relationship, and cohabiting mothers actually report higher levels
of emotional support than do married mothers. White mothers are the least likely to be in an
abusive relationship, and father employment reduces the odds by half that there is abuse in the
relationship. Whites are the most likely to receive substantial emotional support from the baby’s
father, as are women born outside of the US and those who report weekly religious service
attendance. Education, father employment, and mother’s health are also linked with more
emotional support.
Again, comparing cohabitors and visitors to married mothers does not give a complete
picture. Model 3 shows that mothers with low marriage expectations have higher levels of
disagreement, are more likely to be in physically abusive relationships, and receive significantly
less emotional support. Mothers with high marriage expectations report similar relationship
quality as married mothers.
Predicted Probabilities
Tables 6 and 7 present the predicted probabilities of scoring high (as defined in the
previous section) on each of the attitudinal and relationship quality measures, based on the
logistic regression models just discussed, to show the prevalence among the groups.
21
DRAFT 9/2/02
Married versus Unmarried Mothers
As predicted, married mothers are almost twice as likely as unmarried mothers to believe
marriage is best for a couple, all else equal (66 percent versus 38 percent). This is not surprising
given that these women chose to marry. However, the extent to which married mothers’
attitudes changed as a result of marriage is unclear. Looking at marriage for children,
surprisingly, fewer than half of married mothers report that marriage is the best institution for
children, yet, this is still double that of unmarried mothers. Married moms are less likely to think
of marriage as a financial institution (19 percent versus 29 percent), yet again, married mothers
might have become less concerned about the financial benefits of marriage after marriage.
Few mothers report being very traditional or having low levels of trust in men. However,
the differences between married and unmarried mothers are quite large. Fifteen percent of
married mothers are traditional compared to less than 10 percent of unmarried mothers. Six
percent of married mothers report low levels of trust in men, compared to almost nine percent of
unmarried mothers.
Differences in relationship quality, however, are small. Few parents, married or not,
report high levels of disagreement, and less than three percent of mothers report physical abuse.
As stated in the previous section, these reports might be suppressing actual levels of conflict as
they are given just following the birth of a child. Nevertheless, if this is so, married and
unmarried mothers seem to be “underreporting” similarly. Approximately two-fifths of all
mothers report strong emotional support from the baby’s father, and this does not vary much by
marital status.
Cohabitors versus Visitors
22
DRAFT 9/2/02
About 40 percent of cohabiting and visiting mothers think marriage is the best institution
for a couple (37 percent and 40 percent respectively). Even fewer think it is the best institution
for a child; 27 percent of cohabitors and 21 percent of visitors. Cohabiting mothers are slightly
less likely than visitors to think of marriage as a financial institution, yet less than a third of the
mothers in each group scores high on this attitude.
Mothers in cohabiting and visiting relationships have similar attitudes about gender roles,
but cohabiting mothers report better relationship quality than visiting mothers. Very few
mothers report high levels of disagreement, and cohabiting and visiting mothers are equally
likely to be in a relationship that is physically abusive (2.3 percent). However, cohabitors are
much more likely to report that their partner is emotionally supportive. In fact, cohabitors are
more likely than married mothers to report strong emotional, all else equal (48 percent versus 43
percent, respectively).
Marriage Expectations
Table 7 shows the predicted probabilities of scoring high on the eight attitudinal and
relationship quality variables, similar to what was shown in table 6, taking into consideration
marriage expectations. When all cohabitors were compared to all visitors, cohabitors tended to
have more pro marriage attitudes and better relationship quality, however, this picture was
incomplete.
Thirty-nine percent of cohabiting mothers with high marriage expectations think marriage
is the best institution for the couple, compared to 28 percent of cohabiting mothers with low
marriage expectations. The difference for visitors is even larger (47 percent versus 32 percent).
The same pattern holds for considering marriage the best institution for children. Over a quarter
23
DRAFT 9/2/02
of mothers with high marriage expectations are pro marriage for the child compared to 19
percent of cohabitors and 16 percent of visitors with low expectations.
Mothers with high and low expectations differ most in their trust in men, the presence of
physical abuse in the relationship, and in the emotional support the mother feels from the father.
Mothers with low marriage expectations are twice as likely as those with high expectations to
think that men are not trustworthy (13 percent versus about 6 percent), four times as likely to be
abused by the baby’s father, and half as likely to report strong emotional support. Indeed,
mothers with high marriage expectations report similar levels of trust in men and relationship
quality as married mothers do.
CONCLUSION AND POLICY IMPLICATIONS
The results suggest that expectations of marriage are a key indicator of a couple’s
marriage potential. Comparing married parents to unmarried parents, based on living
arrangement alone, provides incomplete information regarding the differences in human capital,
attitudes, and relationship quality. Cohabitors are more similar to married couples than are
visitors, but this is largely explained by differences in marriage expectations. Interacting living
arrangement with marriage intentions shows that high expectations of marriage are a better
predictor than cohabitation of an unmarried mother’s marriage potential.
This paper extended the current research on comparing cohabiting couples to married
couples in three ways. I focused only on new parents, so the results show the differences
between married and unmarried parents, rather than including couples with and without children.
This is important given that it is unmarried relationships that involve children that are of greatest
concern to the public.
24
DRAFT 9/2/02
Secondly, I included romantically involved parents that were not living together at the
child’s birth, but plan to raise their child together (visitors). The results show that including this
group in subsequent analyses is important when considering options to help stabilize unmarried
families. Visitors are similar to cohabitors on most domains, and those with high marriage
expectations are perhaps better candidates for marriage than are cohabitors with low expectations
of marriage.
Finally, by differentiating the unmarried parents based on their living arrangement and
marriage expectations, I showed that parents with higher marriage expectations have higher
human capital, attitudes more favorable for marriage, and better relationship quality than do
couples with low marriage intentions. Marriage expectations are independent of capacity.
Mothers seem to align their expectations of marriage with their attitudes and relationship quality.
Asking a mother if she plans to marry the father of her child may yield a great deal of
information on the quality of the relationship and the couple’s marriage potential.
This is useful information for targeting the right policies toward the right individuals.
These results show that parents with plans to marry seem to be better potential marriage
candidates in terms of relationship quality and attitudes. When differences in capacity are taken
into consideration, mothers with high expectations of marriage report better relationship quality
than do married moms. However, differences in capacity between married and unmarried
parents are huge. Therefore, education and training, employment programs, child care, and
health insurance might be the right solutions for these parent, with only minimal needs for
relationship counseling.
For couples with no plans to marry, they need even greater investments in human capital
development, and they also need intensive relationship counseling, particularly in the areas of
25
DRAFT 9/2/02
increasing their trust and relationship quality. Even if marriage is not the goal for these parents,
they share a child together and learning to cooperate will benefit the child’s future.
A portion of unmarried parents may not be good candidates for marriage at all, including
parents who were not in a romantic relationship when the child was born and those in violent or
high conflict relationships.
Overall, this analysis confirms that high expectations of marriage are an important
indicator of a couple’s marriage potential and must be considered in addition to living
arrangement. Further, the differences among the unmarried couple’s reveal that a one-size-fits-
all type of policy may be ineffective in meeting the diverse needs of unmarried parents.
26
DRAFT 9/2/02
Table 1: Demographic and Human Capital Variables by Relationship Status
Married N=1186
All Unmarried N=3035
Cohabiting N=1774
Visiting N=1261
Mother’s Race White 41.9 13.8 18.1 7.9 Black 24.8 54.7 44.4 69.2
Hispanic 25.4 28.3 34.2 19.9 Other
7.9 3.2 3.3 3.1
Father’s Race White 41.2 11.0 15.1 5.4 Black 26.2 57.3 46.8 72.0
Hispanic 24.4 28.5 34.7 19.7 Other 8.2 3.2 3.4 2.9
Mother Foreign Born 24.5 13.7 17.4 9.9 Age
Mother’s Age 29.3 23.9 24.2 23.5 Father’s Age 31.7 26.8 27.1 26.4
Mother Attends Weekly Religious Service 33.5 18.2 15.2 20.1
Employment
Mother 71.4 66.0 68.9 61.9 Father* 91.6 74.9 80.7 66.8
Education Mother’s
Less than H.S. 16.6 40.5 40.1 40.9 High School 19.9 33.7 33.4 34.2
Some College 28.8 22.8 23.3 22.0 College 34.6 3.1 3.2 2.9
Father’s* Less than H.S. 16.9 39.7 39.8 39.6
High School 24.6 37.5 35.0 40.9 Some College 27.7 19.8 21.9 16.9
College 30.7 2.9 3.2 2.5 Medicaid Paid for Birth 26.8 73.4 71.7 75.4 Health Mother’s
Excellent 73.4 63.3 62.1 64.9 Good 22.2 28.2 29.1 26.9 Poor 4.4 8.5 8.8 8.2
Father’s Health Limits Work 3.2 6.3 6.8 5.7 High Marriage Expectations NA 66.2 78.4 49.1
*Father’s employment and education are based on father’s report where available, and supplemented with mother’s report.
27
DRAFT 9/2/02
Table 2: Demographic and Human Capital Variables by Relationship Status and Expectations of Marriage
Married N=1186
Cohabiting/ High Chance of
Marriage N=1390
Cohabiting/ Low Chance of
Marriage N=384
Visiting/ High Chance of
Marriage N=619
Visiting/ Low Chance of
Marriage N=642
Mother’s Race White 41.9 20.4 9.9 8.7 7.0 Black 24.8 41.9 53.4 68.0 70.3
Hispanic 25.4 34.4 33.6 20.0 19.8 Other
7.9 3.4 3.1 3.2 2.9
Father’s Race White 41.2 17.2 7.3 7.1 3.7 Black 26.2 44.5 55.5 70.1 73.8
Hispanic 24.4 34.8 34.6 20.0 19.3 Other 8.2 3.6 2.6 2.8 3.1
Mother Foreign Born 24.5 17.1 18.5 9.1 10.9 Age
Mother’s Age 29.3 24.2 24.4 23.1 23.9 Father’s Age 31.7 26.9 27.7 25.7 27.0
Mother Attends Weekly Religious Service 33.5 16.4 10.9 21.7 18.7
Employment
Mother 71.4 69.9 65.6 65.1 58.9 Father* 91.4 81.9 75.8 70.3 63.2
Education Mother’s
Less than H.S. 16.6 38.1 47.7 36.2 45.5 High School 19.9 34.3 29.9 36.7 31.8
Some College 28.8 24.0 20.6 24.2 19.9 College 34.6 3.6 1.8 2.9 2.8
Father’s* Less than H.S. 16.9 38.3 45.3 35.1 43.9
High School 24.6 35.0 34.9 44.1 38.0 Some College 27.7 23.4 16.4 18.7 15.3
College 30.7 3.3 3.4 2.1 2.8 Medicaid Paid for Birth 26.8 69.3 80.5 74.8 76.0 Health Mother’s
Excellent 73.4 64.0 55.2 66.4 63.4 Good 22.2 27.8 33.6 27.1 26.8 Poor 4.4 8.1 11.2 6.5 9.8
Father Health Limits Work 3.2 6.3 8.6 4.0 7.3 *Father’s employment and education are based on father’s report where available, and supplemented with mother’s report.
28
DRAFT 9/2/02
Table 3: Results from Logistic Regressions of Attitudes about Marriage Odds Ratios
Pro Marriage for the
Couple (Odds of Scoring 6 out of 8)
Pro Marriage for the Child’s Well-Being
(Odds of Scoring 6 out of 8)
Marriage as a Financial Institution
(Odds of Scoring 9 out of 9)
(1) (2) (3) (1) (2) (3) (1) (2) (3) Relationship Status
(Married) Cohabiting .24*** .30*** .31*** .46*** 2.54*** 1.59***
Cohabiting/High .33*** .50*** 1.61*** Cohabiting/Low .19*** .29*** 1.51***
Visiting .32*** .35*** .21*** .33*** 3.29*** 1.88*** Visiting/High .46*** .42*** 2.00*** Visiting/Low .25*** .24*** 1.76***
Demographic Characteristics
Age Mother 1.00 1.00 1.00 1.00 1.00 1.00 Foreign Born 1.04 1.05 1.39*** 1.40*** 1.08 1.09 Race Mother
(Black)
White .61*** .59*** 1.61*** 1.57*** .37*** .37*** Hispanic .49*** .48*** 1.43*** 1.40*** .94 .94
Other .75* .73* 2.01*** 1.98*** .80 .80 Weekly Religion 2.09*** 2.06*** 1.78*** 1.75*** 1.09 1.09
Human Capital Education Mother
(< High School)
High School 1.08 1.05 .87 .84* .94 .94 Some College 1.32*** 1.28*** .95 .93 .62*** .62***
College 1.75*** 1.70*** 1.37** 1.35** .33*** .32*** Employment
Mother .99 .99 .84** .84** 1.21*** 1.21*** Father 1.19** 1.17* 1.01 .99 .86 .89
Medicaid .98 .99 .88 .89 1.07 1.07 Health Mother
Great 1.09 1.06 1.03 .99 .85 .85 Good 1.16 1.13 1.08 1.06 .90 .89
(Poor) N 4221 4221 4221 4221 4221 4221 4221 4221 4221 R-squared .0599 .0949 .1021 .0683 .0951 .1004 .0331 .0686 .0689
*** P value less than or equal to .01 ** P value less than or equal to .05 * P value less than or equal to .10
29
DRAFT 9/2/02
Table 4: Results from Logistic Regressions of Attitudes about Gender Roles Odds Ratios
Traditional Gender Roles (Odds of Scoring 6 out of 8)
Gender Distrust (Odds of Scoring 6 out of 8)
(1) (2) (3) (1) (2) (3) Relationship Status
(Married) Cohabiting .69*** .63*** 1.76*** 1.29*
Cohabiting/High .62*** 1.08 Cohabiting/Low .63*** 2.26***
Visiting .56*** .53*** 2.00*** 1.55*** Visiting/High .58*** .96 Visiting/Low .48*** 2.24***
Demographic Characteristics
Age Mother 1.05*** 1.05*** 1.05*** 1.05*** Foreign Born 1.14 1.14 1.48*** 1.47*** Race Mother
(Black)
White .87 .87 .81 .87 Hispanic 1.62*** 1.62*** 1.35** 1.42***
Other 1.61** 1.61** 1.68** 1.76** Weekly Religion 2.08*** 2.07*** 1.08 1.14
Human Capital Education Mother
(< High School)
High School .54*** .54*** .64*** .66*** Some College .39*** .39*** .41*** .42***
College .34*** .34*** .23*** .23*** Employment
Mother .72*** .71*** .68*** .68*** Father 1.07 1.07 .82 .84
Medicaid 1.16 1.16 1.37*** 1.34** Health Mother
Great .75* .74* .48*** .49*** Good .93 .92 .65*** .66***
(Poor) N 4221 4221 4221 4221 4221 4221 R-squared .0077 .0836 .0839 .0095 .0881 .1021
*** P value less than or equal to .01 ** P value less than or equal to .05 * P value less than or equal to .10
30
DRAFT 9/2/02
Table 5: Results from Logistic Regressions of Relationship Quality Odds Ratios
Disagreement (Odds of Scoring < =9 out of 18)
Physical Abuse (Hit/Slap Sometimes or Often)
Supportiveness (Odds of Scoring 9 out of 9)
(1) (2) (3) (1) (2) (3) (1) (2) (3) Relationship Status
(Married) Cohabiting 2.10* 1.57 1.97*** 1.22 .94 1.21**
Cohabiting/High 1.27 .73 1.48*** Cohabiting/Low 2.70** 3.10*** .46***
Visiting 4.58*** 2.95*** 2.27*** 1.23 .50*** .68*** Visiting/High 1.89 .48 1.17 Visiting/Low 4.08*** 2.06*** .32***
Demographic Characteristics
Age Mother 1.05*** 1.05** 1.03 1.02 .98*** .98*** Foreign Born .44* .43* .94 .89 1.18* 1.22** Race Mother
(Black)
White .49 .52 .34*** .38** 1.39*** 1.32*** Hispanic .79 .81 1.14 1.23 1.08 1.03
Other 1.28 1.34 2.14** 2.31** 1.07 1.02 Weekly Religion .61 .63 .68 .74 1.35*** 1.29***
Human Capital Education Mother
(< High School)
High School 1.14 1.19 .82 .89 1.09 1.02 Some College .99 1.05 .71 .78 1.12 1.05
College .42 .44 .63 .68 1.58*** 1.50*** Employment
Mother .71 .72 .92 .95 1.07 1.06 Father .62* .64* .52*** .54*** 1.36*** 1.30***
Medicaid .83 .81 1.63** 1.56* .98 1.02 Health Mother
Great .37*** .38*** .77 .82 1.52*** 1.44*** Good .59 .59 .65 .67 1.14 1.06
(Poor) N 4221 4221 4221 4221 4221 4221 4221 4221 4221 R-squared .0294 .0748 .0855 .0096 .0512 .0907 .0154 .0339 .0674
*** P value less than or equal to .01 ** P value less than or equal to .05 * P value less than or equal to .10
31
DRAFT 9/2/02
Table 6: Predicted Probabilities of Attitudes About Marriage, Gender Roles, and
Relationship Quality by Relationship Status Married
N=1186 All Unmarried
N=3035 Cohabiting
N=1774 Visiting N=1261
Attitudes About Marriage Pro Marriage for Couple
65.5 37.8 36.6 39.8
Pro Marriage for Child’s Well-Being
44.1 24.3 26.6 20.8
Marriage as a Financial Institution
19.1 28.6 27.3 30.7
Attitudes About Gender Roles
Traditional Gender Roles
15.1 9.5 10.0 8.6
Gender Distrust
6.3 8.6 8.0 9.4
Relationship Quality Often Disagree
0.7 1.4 1.1 2.0
Physically Abusive
1.9 2.3 2.3 2.3
Often Supportive
43.2 42.5 47.8 34.2
____________ Predicted probabilities of those who agree or strongly agree are listed, except for relationship quality which is noted. Mother’s race, nativity, age, religiosity, employment, education, Medicaid use and health, and father’s employment are controlled at their means.
32
DRAFT 9/2/02
Table 7: Predicted Probabilities of Attitudes About Marriage, Gender Roles, and Relationship Quality by Relationship Status and Expectations of Marriage
Married N=1186
Cohabiting/ High Chance of
Marriage N=1390
Cohabiting/ Low Chance of
Marriage N=384
Visiting/ High Chance of
Marriage N=619
Visiting/ Low Chance of
Marriage N=642
Attitudes About Marriage
Pro Marriage for Couple
65.8 38.9 27.7 46.9 32.4
Pro Marriage for Child’s Well-Being
44.4 28.7 18.7 25.2 16.4
Marriage as a Financial Institution
19.1 27.5 26.2 32.1 29.3
Attitudes About Gender Roles
Traditional Gender Roles
15.1 9.9 10.1 9.4 7.8
Gender Distrust
6.2 6.7 13.0 6.0 12.9
Relationship Quality
Often Disagree
0.7 0.9 1.8 1.3 2.7
Physically Abusive
1.9 1.4 5.6 0.9 3.8
Often Supportive
43.8 53.6 26.2 47.8 19.9
____________ Predicted probabilities of those who agree or strongly agree are listed, except for relationship quality, which is noted. Mother’s race, nativity, age, religiosity, employment, education, Medicaid use and health, and father’s employment are controlled at their means.
33
DRAFT 9/2/02
Reference List Brown, Susan. 2000. “Union transitions among cohabitors: the significance of relationship
assessments and expectations.” Journal of Marriage and the Family 62: 833-846. Brown, Susan, and Alan Booth. 1996. “Cohabitation versus marriage: a comparison of
relationship quality.” Journal of Marriage and the Family 58: 668-678. Bumpass, Larry and Hsien-Hen Lu. 1999. “Trends in cohabitation and implications for
children’s family context in the U.S.” Working paper No. 98-15, Center for Demography and Ecology, University of Wisconsin, Madison.
Bumpass, Larry, James Sweet, and Andrew Cherlin. 1991. “The role of cohabitation in
declining rates of marriage.” Journal of Marriage and the Family 53:913-927. Carlson, Marcia, Sara McLanahan, and Paula England. 2002. “Union formation and stability in
Fragile Families.” Center for Research on Child Wellbeing, Working Paper #01-06-FF, Princeton, NJ.
Casper, Lynn and Liana Sayer. 2000. “Cohabitation transitions: different attitudes and purposes,
different paths.” Paper presented at the annual meeting of the Population Association of America.
Clarkberg, Marin, Ross Stolzenberg, and Linda Waite. 1995. “Attitudes, values, and entrance
into cohabitational versus marital unions.” Social Forces 74: 609-634. Duncan, Greg and Willard Rodgers. 1991. “Has children’s poverty become more persistent?”
American Sociological Review 56: 538-550. Graefe, Deborah Roempke and Daniel Lichter. 1999. “Life course transitions of American
children: parental cohabitation, marriage, and single motherhood.” Demography 36: 205-217.
Manning, Wendy and Daniel Lichter. 1996. “Parental cohabitation and children’s economic
well-being.” Journal of Marriage and the Family 58: 998-1010. Manning, Wendy and Pamela Smock. 1995. “Why marry? Race and the transition to marriage
among cohabitors.” Demogaphy 32: 509-520. McLanahan, Sara and Larry Bumpass. 1988. “Intergenerational consequences of family
disruption.” American Journal of Sociology 94:130-152. McLanahan, Sara, and Gary Sandefur. 1994. “Growing Up With a Single Parent: What Helps,
What Hurts.” Cambridge: Harvard University Press.
34
DRAFT 9/2/02
National Center for Health Statistics, 2000. Nonmarital childbearing according to detailed race of mother.
Nock, Steven. 1995. “A comparison of marriages and cohabiting relationships.” Journal of
Family Issues 16: 53-76. Rindfuss, Ronald, and Audrey VandenHeuvel. 1990. “Cohabitation: a precursor to marriage or
an alternative to being single?” Population and Development Review 16: 703-726. Sigle-Rushton, Wendy and Sara McLanahan. 2002. “For Richer or Poorer.” Working paper,
Center for Research on Child Wellbeing, Princeton, NJ. Smock, Pamela. 2000. “Cohabitation in the United States: an appraisal of research themes,
findings, and implications.” Annual Review of Sociology 26: 1-20. Smock, Pamela, and Wendy Manning. 1997. “Cohabiting partners’ economic circumstances and
marriage.” Demography, 34: 331-341. Waller, Maureen, and Sara McLanahan. 2001. “Do unmarried parents’ expectations predict
marital transitions? Early evidence from the Fragile Families and Child Well-Being study.” Paper presented at the annual meeting of the American Sociological Association.
35